Langenskiold classification of tibia vara: a multicenter study on interrater reliability

Since 1952, when Langenskiold introduced his six-stage classification on roentgenographic morphologies of tibia vara, this method has been used extensively in Orthopaedic studies. Different studies have tried to determine the reliability of this method. Although there are scarce reports of major variability in using Langenskiold classification, but the question still remains, as to whether using expert subspecialists and MRI would improve its reliability. The aim of this study is to evaluate inter-rater reliability of Langenskiold classification using simple radiograph and MRI, by expert raters. This is a cross-sectional multicenter study, involving patients from 4 tertiary pediatric orthopedic centers in France and Iran. Radiograph and MRI (T1, T2, fat saturation) of the affected knees were independently classified by 12 academy members from France and Iran: six pediatric fellowship orthopedic surgeons and six musculoskeletal radiologists. All data were analyzed afterwards by an independent researcher. Mean weighted kappa for agreement based on radiograph and MRI was 0.47 and 0.45, respectively. Mean percent of total agreement was 46.6 and 40.8% in the same order. Kappa statistics for the new grouping were 0.53 and 0.46 for roentgenograms and MRIs, respectively. The new regrouping was proposed so that each group would need a different treatment strategy. Fleiss kappa statistics for group B (stages 2 and 3), rose from (0.33 and 0.09) to 0.44. These values for group C (stages 4 and 5) rose from (0.42 and 0.11) to 0.54. This study concludes that Langenskiold's classification when used by experts, still has a moderate inter-rater reliability at best. This variability is most at stages 2, 3, and 4; and can cause different treatment approaches. Use of MRI does not have a significant effect on its reliability. Regrouping the stages improved the inter-rater reliability. Level of evidence: III.

Automated summary

Langenskiold classification, widely used in Orthopaedic studies, has moderate inter-rater reliability. Variability is most prominent at stages 2, 3, and 4, leading to different treatment approaches. The use of MRI does not significantly improve its reliability. Regrouping the stages can enhance inter-rater reliability.